These test cases cover obfuscation (more properly, masking) of literal values from SQL statements captured by agents. SQL statements may be captured and attached to transaction trace nodes, or to slow SQL traces.
sql_obfuscation.json contains an array of test cases. The inputs for each
test case are in the sql property of each object. Each test case also has an
obfuscated property which is an array containing at least one valid output.
Test cases also have a dialects property, which is an array of strings which
specify which sql dialects the test should apply to. See "SQL Syntax Documentation" list below. This is relevant because for example, PostgreSQL uses
different identifier and string quoting rules than MySQL (most notably,
double-quoted string literals are not allowed in PostgreSQL, where
double-quotes are instead used around identifiers).
Test cases may also contain the following properties:
malformed: (boolean) tests who's SQL queries are not valid SQL in any quoting mode. Some agents may choose to attempt to obfuscate these cases, and others may instead just replace the query entirely with a placeholder message.pathological: (boolean) tests which are designed specifically to break specific methods of obfuscation, or contain patterns that are known to be difficult to handle correctlycomments: an array of strings that could be usefult for understanding the test.
The following database documentation may be helpful in understanding these test cases:
SQL Syntax Documentation: